Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Combined Load Test and NDE: Novel Method to Diagnose

and Load Rate Prestressed Concrete Girder Bridges


with Serviceability Issues
Ikram Efaz, Ph.D., M.ASCE1; Nur Yazdani, Ph.D., P.E., F.ASCE2;
and Eyosias Beneberu, Ph.D., P.E., M.ASCE3
Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Vibration of highway bridges, although often ignored, is a serviceability issue that may contribute to concrete deck cracking and
delamination. A recently constructed prestressed concrete bridge in the Dallas–Fort metroplex experienced extensive transverse deck cracks,
potholes, and excessive vibration. The bridge was load-rated employing a novel and combined Nondestructive Evaluation (NDE) and load
test method. Ground Penetrating Radar scans showed satisfactory concrete cover for the top reinforcement in the deck, indicating negligible
surface wearing and construction error. Impact Echo scans of the top of the deck revealed severe concrete delamination at the top of the gird-
ers with fully cast-in-place (CIP) concrete and also at the interface of the precast deck panels and the CIP topping. The calculated neutral axis
depths of the composite girders using an initial and follow-up load test after about a year were found to be in the web, indicating only 59%
partially composite action. It was concluded that the excessive vibration was due to the deck delamination and the partial-composite action
that reduced the stiffness of the superstructure. Both the deck and the composite girders can safely carry HS-20 live loading as required by the
Texas Department of Transportation. However, the long-term bridge serviceability and durability could be compromised due to the concrete
degradation and loss of stiffness. DOI: 10.1061/(ASCE)BE.1943-5592.0001939. © 2022 American Society of Civil Engineers.
Author keywords: Bridge vibration; Serviceability; Combined Nondestructive Evaluation and load test; Concrete delamination; Partial-
composite action; Load factor method.

Background correlation between the road surface roughness and the dynamic
impact factor of bridges through full-scale dynamic testing, while
According to the National Bridge Inventory (2017), out of another study by Kou and Dewolf (1997) reinforced the finding.
617,000 bridges in the United States, 46,154 are structurally defi- Awall et al. (2012) concluded that expansion joint defects greatly
cient. Almost 4 in 10 bridges are more than 50 years old, which impact bridge vibration and suggested frequent maintenance to re-
means an increasing number of bridges will need major rehabili- duce expansion joint bump heights.
tation soon (ASCE 2021). However, serviceability issues of these Alampalli (2001) concluded that vibration severity is the
bridges are often overlooked. For example, the AASHTO LRFD most significant parameter influencing bridge deck cracking,
Bridge Design Specifications (AASHTO 2017) contain only op- particularly for long-span bridges. Like other types of cracking,
tional deflection limits to control the vibration of highway brid- vibration-induced deck cracking and delamination can be inspected
ges. However, last few studies on the vibration parameters of and mapped using several Nondestructive Evaluation (NDE) tech-
bridges, such as deflection, acceleration, velocity, and frequency, niques. Zaki et al. (2018) successfully used Ground Penetrating
found that AASHTO deflection limits are inadequate to control vi- Radar (GPR) to assess the presence of rebar corrosion in a concrete
bration (Roeder et al. 2002; Barker and Barth 2013; Barker and slab. Similarly, Raju et al. (2018) and Hasan and Yazdani (2015)
Staebler 2011; Gaunt and Sutton 1981; Fountain and Thunman quantitively related concrete rebar corrosion with the maximum re-
1987). flected wave amplitude from GPR data. Hasan and Yazdani (2014)
Vibration causes an increase in stresses, which is accounted for also used GPR to evaluate reinforcement clear cover variation in a
by the dynamic amplification factor (AASHTO 2017). Several fac- new bridge deck and found that a significant portion (48%) had in-
tors affect the vibration response of bridges, such as span length, adequate cover. It is evident that less concrete cover may result in
stiffness, and vehicle speed. Park et al. (2005) showed a strong corrosion of reinforcement and excess cover tends to reduce mo-
ment capacity due to a reduced moment arm. Apart from GPR, Im-
1
E.I.T., Tower Engineering Solutions, LLC, 1320 Greenway Dr., pact Echo (IE) can be used to evaluate and detect possible
Irving, TX 75038. delamination in concrete bridge decks. Gucunski et al. (2005)
2
Dept. of Civil Engineering, Univ. of Texas at Arlington, P. O. Box used IE and GPR on asphalt-overlaid bridge decks to estimate con-
19308, Arlington, TX 76019-0308. crete deterioration and possible delamination. Similarly, Kee et al.
3
Bridgefarmer & Associates, Inc., 2350 Valley View Lane, Dallas, TX (2012) utilized air-coupled IE and Infrared Thermography (IR) on a
75234 (corresponding author). ORCID: https://orcid.org/0000-0002-8122 reinforced concrete bridge deck containing simulated delamination
-1985. Email: eyosias.beneberu@mavs.uta.edu
and cracking defects. The results from both the NDE methods
Note. This manuscript was submitted on February 8, 2022; approved on
June 7, 2022; published online on August 4, 2022. Discussion period open showed good agreement with most of the actual defects. Gucunski
until January 4, 2023; separate discussions must be submitted for individual et al. (2008, 2009) proposed a new automated approach based on
papers. This paper is part of the Journal of Bridge Engineering, © ASCE, three-dimensional IE data, which effectively enabled IE to be
ISSN 1084-0702. used as a bridge deck sonar device.

© ASCE 04022097-1 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


NDE methods like GPR and IE can locate localized cracks Department of Transportation’s (TxDOT) local area office that
and delamination in the superstructure, while in situ load testing has jurisdiction over this bridge expressed concern about the exces-
can accurately determine bridge load-carrying capacity and sive vibrations. Therefore, upon the request of the TxDOT local
provide a better understanding of the load paths and their distri- area office, a study team from the University of Texas at Arlington
bution through the bridge components (Bujnakova et al. 2018; initially visited the bridge site on January 29, 2018, and conducted
Lichtenstein 1995). Diagnostic load tests are performed to deter- visual inspections. Extensive transverse and longitudinal cracks
mine bridge characteristics, load response, and distribution or to were observed on top of the deck, together with a few patched pot-
validate analytical models (AASHTO 2017). Several Depart- holes. The SB bridge with four spans and carrying five lanes of traf-
ment of Transportations (DOTs) across the United States have fic was selected for this study. Specifically, span 3 was chosen for
been using diagnostic load testing effectively to evaluate condi- evaluation since this span was observed to have the most number of
tion assessment and load rating of existing bridges (Almomani cracks and vibration intensity. The deck thickness was 216 mm,
et al. 2020; Lucas et al. 2004; Hag-Elsafi and Kunin 2006; consisting of composite cast-in-place (CIP) deck on precast pre-
Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

Matta et al. 2005; Bettis 2020). stressed panels. The total width of the deck was 29.87 m measured
A newly constructed prestressed concrete girder bridge in the from outside of the traffic barriers. The bridge has 11 girders at a
Dallas-Fort Worth area was observed to have progressive crack- spacing ranging from 2.74 to 2.9 m. The plan view, typical section
ing and localized potholes on the deck due to excessive vibration of TX54 prestressed girders, deck cross section with girder label-
of the superstructure. Although it is likely that the existing com- ing, and precast prestressed deck panel details in span 3 are
posite action between the deck and the girders in most concrete shown in Fig. 2. The relevant material properties for span 3 per as-
bridges has never been 100% effective due to construction or de- built drawing are given in Table 1.
terioration issues, the relationship between possible partial deck- Extensive transverse and longitudinal cracks were observed on
girder composite action and excessive vibration in a concrete top of the deck as shown by lighter shaded lines in Fig. 3(a). The
bridge has never been established to date. The specific objectives Texas Department of Transportation (TxDOT) officials informed
of this research were to: (1) Investigate the cause of bridge the authors that a pothole appeared on top of the deck and was sub-
vibration using a diagnostic load test and NDE; (2) Evaluate sequently repaired as shown in Fig. 3(b). In addition to the ob-
any progressive deterioration by conducting a follow-up diagnos- served deterioration, excessive vibration was also felt on the deck
tic load test; and (3) Propose a novel technique of load rating of as heavy traffic passed over the bridge. Because it is a fairly new
the bridge girders and the deck combining the NDE and load bridge, it was unusual for it to experience excessive vibration,
test data. No similar research was conducted in the past to learn cracking, and potholes.
the cause of vibration for a newly constructed concrete bridge
using this approach. The current investigation sets a benchmark
that could eventually be used as a reference for future studies re-
Nondestructive Evaluation Procedure
garding prestressed concrete girder bridges with vibration issues
and their effective load rating.
Truck-mounted GPR with two 2.6 GHz and one 400 MHz anten-
nae were used to scan the bridge deck at a speed of about 2.24 m/s
to determine the reinforcement cover and location as shown in
Description of the Test Bridge Fig. 4(a). The first antenna has a depth range of 0–305 mm,
while the latter has a depth range of 0–3.7 m. Both antennae
Built in 2011, the State Highway 75 (SH 75) Bridge over Wilson were used to cover multiple depths at the same time. This antenna
Creek in McKinney, Texas, has both North Bound (NB) and configuration saves time compared with the usual method of GPR
South Bound (SB) bridges, as shown in Fig. 1. The Texas scanning with a tri-wheel pushcart using a single antenna. The IE

(b)

(a)

Fig. 1. SH-75 SB Bridge over Wilson Creek Bridge, McKinney, Texas: (a) location (map data © 2020 Google, INEGI); and (b) street view (image ©
2018 Google).

© ASCE 04022097-2 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

(c)

(d)

Fig. 2. Span 3 of the SB Willson Creek Bridge: (a) plan view; (b) cross section of the TX54 girder; (c) transverse section; and (d) precast prestressed
deck panel details.

Table 1. Properties of concrete and steel system, shown in Fig. 4(b), was used to detect delamination in the
deck due to voids, honeycombs, and/or cracks.
Properties Value To facilitate the NDE work with an appropriate traffic lane clo-

Deck concrete compressive strength ( fc,deck ) 28 MPa sure, scanning was performed over three days. Lanes 1, 2, and 3

Girder concrete compressive strength ( fc,girder ) 43 MPa and the west side shoulder, as shown in Fig. 4(c), were scanned
Number of prestressing strands in each TX54 girder 50 on the first two days, and the remaining Lanes 4 and 5 and the
Strand size 13 mm east side shoulder were scanned on the third day. The IE data
Strand ultimate strength 1,862 MPa
were collected in a grid pattern, as shown in Fig. 5, by spot

© ASCE 04022097-3 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

Fig. 3. Deterioration and repair on the deck: (a) visible cracks; and (b) repaired pothole.

(a) (b)

(c)

Fig. 4. NDE devices: (a) truck-mounted GPR; (b) Impact Echo (IE) (images by Ikram Efaz); and (c) lane configuration.

scanning at spacings ranging from 0.76 to 1.52 m, while the GPR superstructure behavior was investigated. A total of 26 strain
data were collected at a constant spacing of 1.52 m. gauges [Fig. 6(b)] and 12 rotational tiltmeters [Fig. 6(c)] were in-
stalled on the girders to monitor strain and rotation, respectively.
Some girders were equipped with both top and bottom strain
Load Test Procedure gauges to find the location of the neutral axis. The rotational tiltme-
ters were installed on six selected girders at 0.61 m from the center
A detailed plan was prepared for the instrumentation and static load of the bents. All the sensors were installed using a boom lift. The
testing of Span 3 of the SB Bridge, as shown in Fig. 6(a). Only the installation procedure involved marking the mounting points,

© ASCE 04022097-4 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Working grid for IE scanning.

cleaning the concrete surface using acetone, and attaching the sen- To evaluate any time-dependent rate of deck deterioration, a
sors using an adhesive. The sensor installation did not require lane follow-up load test was conducted on October 7, 2019. An instru-
closure since there was plenty of vehicular traffic-free space under mentation scheme similar to the first load test was employed, ex-
the bridge to move the boom lift and provide a safe working envi- cept that six additional strain gauges were installed on the deck
ronment. DS-50A and STS4 DAQ boxes were used to collect the bottom. To obtain comparable data, trucks similar to the initial
data from the instrumentation at a rate of 10 data per second. load test were used [Fig. 6(d), Table 2]. However, the gross vehicle
The load testing was conducted on May 24, 2018, using three weights of Trucks A and C from the initial load test were slightly
trucks (A, B, and C), with weights shown in Table 2. The distance higher than the follow-up test. In addition, a procedure in terms
between the front and the middle axles was 4.11 m, while the dis- of load paths and speed of vehicles similar to the initial load test
tance between the middle and the rear axles was 1.37 m, as shown was employed during the follow-up test.
in Fig. 6(d). The trucks were provided by the TxDOT Dallas Dis-
trict and were filled with a dense material to yield the maximum
possible axle weights. Load Rating Methodology
The beginning, middle, and end of the span, together with var-
ious paths for different truck runs, were marked on the day of the The data from GPR, IE, and load tests were employed for the load
load testing. Fig. 7(a) shows different paths, while Fig. 7(b) rating of the bridge. The first two were used for the deck, while the
shows Truck C in position for the crawl speed test as per the last was used for the girders. Fig. 8 shows a flow chart demonstrat-
AASHTO Manual for Bridge Evaluation (AASHTO 2018). The ing the proposed load rating methodology. The 2-D GPR contour
truck started before Bent 4 and stopped when it fully exited Bent plot for the top reinforcement was divided into negative and pos-
3. Two runs with different speeds were conducted along Paths itive moment regions, considering the contraflexure points of a
P1, P2, and P3 to verify the repeatability of the results. As continuous deck. Then, the negative moment regions were ana-
shown in Fig. 7(a), Path P1 was selected to produce maximum ef- lyzed in an image processing software (Image Color Extract
fect on the exterior Girder B1., while Paths P2 and P3 were selected 2016) to extract the percentage of different colors corresponding
to find the effects on the interior Girders B2 to B10. The path de- to a range of values of the top mat reinforcement cover. The aver-
tails, speed, and truck wheel locations for each test are given in age reinforcement cover obtained from this process was used to
Table 3. A stop location test, as shown in Fig. 7(c), was also per- calculate the moment arm for the negative moment capacity of
formed using three trucks moving side by side in Lanes 2, 3, and the deck. Similarly, the 2-D IE contour map was divided into neg-
4 to maximize the effect on the interior girders. The trucks were ative and positive moment regions and analyzed in the image pro-
moved to the midspan and stopped for about 20 s to obtain the re- cessing tool (Image Color Extract 2016). The percentages of light
sponse of the bridge. shade indicating the presence of severe delamination in these two

© ASCE 04022097-5 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

(b)

(a)

(c) (d)

Fig. 6. Load test plan: (a) instrumentation plan; (b) attached strain gauge; (c) attached rotational tiltmeter; and (d) dimensions of the test trucks.

Table 2. Truck axle weights


Truck A (kN) Truck B (kN) Truck C (kN)

Gross and axle weights Initial test Follow-up test Initial test Follow-up test Initial test Follow-up test
Gross vehicle weight (GVW) 191 189 197 197 209 200
Axle 1 42 43 43 50 47 48
Axle 2 74 73 77 74 81 76
Axle 3 74 73 77 74 81 76

regions were then obtained. The deck’s calculated positive and load testing strain data was employed in the capacity calculation
negative flexural capacities were reduced by the average of the instead of the theoretical value. The percentage composite action
percentage delamination of the positive and negative regions. Fi- of the deck-girder system was estimated by comparing the NA
nally, the rating factors for the two regions were calculated, and from the experiment with the theoretically calculated values as-
the smallest of the two values governed the deck rating. For the suming full- and noncomposite actions. The overall bridge rating
girder’s rating, the neutral axis (NA) depth calculated using is the minimum rating of the girder and deck. The proposed

© ASCE 04022097-6 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

(b)

(a) (c)

Fig. 7. Load tests: (a) locations of the trucks; (b) crawl speed test; and (c) stop location test.

Table 3. Truck speeds and locations for the crawl speed test NDE and Load Test
Location of the right wheel from the The GPR scans were post-processed and analyzed using RADAN
Path Run Speed (m/s) west edge of the deck (m) (2018) and GPR-SLICE (2018) software. Two-dimensional
P1 1 2.2 1.4 contour plots of the top reinforcement concrete covers found using
2 the 2.6 GHz GPR antenna are presented in Fig. 9(a). The cover
P2 1 7.3 depth found from the GPR was mostly in the range of 66–71 mm.
2 The as-built drawing specified a cover of 63.5 mm for the top mat
P3 1 21.2 reinforcement indicating that the rebars were placed with minimal
construction error and the deck has experienced negligible surface
wearing. As shown in Fig. 9(a), no concrete defects could be ob-
methodology is unique and more accurate than the industry-wide served from the GPR scans. One possible reason is that the cracks,
bridge load rating approach for the following reasons: (1) It used voids, or delamination should be at least 6.35 mm wide to be visible
the actual constructed top reinforcement clear cover for the deck in the B-scans from the GPR (Goulias and Adojam 2015).
instead of the as-built drawing prescribed value; (2) it considered The IE results on discrete data grids were plotted in 2-D contour
the actual inner core condition of the deck rather than only the maps, as shown in Fig. 9(b). Light shade indicates a high probability
wear and tear on the exposed riding surface of the deck; and (3) of delamination with 3% estimation accuracy (Impact-Echo Instru-
it considered partial deck-girder composite action, unlike the ex- ments 2001), while the dark shaded areas indicate sound concrete
isting prevalent practice of considering either full or no-composite without any debonding or delamination. Contours from both segments
action. show strong signs of delamination through the light shaded areas. The
delamination is more prevalent on the top of the girders compared with
the areas between the girders. Because most of the IE signals were due
Results and Discussion to large-amplitude and low-frequency flexural vibration, the delamina-
tion depths were expected to be within 102 mm from the top surface.
The following sections present results and discussions from the The phenomenon of flexural vibration is frequently encountered on
load tests and the NDE of the SB bridge. shallow flaws and delamination on concrete bridge decks. The IE

© ASCE 04022097-7 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Flowchart of the load rating method.

device cannot specify the exact delamination depth; it only shows the Fig. 10(e) shows the strain versus time diagram near the midspan
potential presence of delamination below the minimum depth limit of for the stop location test. Girders B4, B5, and B6 show the maximum
102 mm (Impact-Echo Instruments 2001). response as they were located near the position of the three trucks.
The strain gauges and the rotational tiltmeters’ data from the As can be seen from the trends in all the plots in Fig. 10, the strain
load test were analyzed to evaluate the load response of the bridge. values started from zero, then increased to maximum values when
The raw strain data exhibited noise due to traffic-induced vibration. the trucks were at the midspan but did not return to zero after the
This was eliminated by calculating the moving average between 10 trucks left the span, leaving some residual strains. Also, some incon-
data points, providing one data point for every second which was sistencies were observed between the two runs for all the paths. For
sufficiently accurate for the evaluation purposes. This common example, the strain gauges for Fig. 10(a) show a slightly higher strain
technique of data smoothening removed the unwanted fluctuations response than Fig. 10(b). The presence of residual strain and incon-
(sharp peaks and dips) in the strain responses and provided filtered sistent results between two different runs could be due to some un-
data. The nomenclature used in labeling the strain gauges is as fol- intentional elastic slip of the foil strain gauges.
lows: the first number denotes the section, the second letter and To evaluate the existing composite action between the deck and
third numbers denote the girder number (B1, B2, and so on), and the girders, the actual NA depths of the girders were calculated
the last letter denotes the location of the strain gauge (B: girder bot- using Eq. (1) from the measured top and bottom strain data. The
tom; T: girder top). Figs. 10(a and b) show the strain-time history calculation took into account assumptions of pure bending, such
from Path P1 near the midspan for Runs 1 and 2. The strain re- as (1) Linear strain profile since the girders were designed to be
sponses from Girders B1 and B2 were maximum as expected within the elastic limit; (2) material homogeneity; and (3) plane
since the truck tires were on top of these girders. The strain across section remains plane after bending.
other girders gradually decreased with an increase in distance from εB d
the truck tires. Because Run 2 was a repetition of Run 1 from Path y = (1)
εB + εT
P1, Figs. 10(a and b) yielded identical results. Figs. 10(c and d)
show the strain versus time plots for Path P2 Run 1 and Path P3 where y = depth of the neutral axis from the bottom (mm); d = dis-
Run 1 near the midspan, respectively. The trends are similar to tance between the top and the bottom gauges (mm); ɛB = strain in
Path P1 since Girders B4 and B7 showed maximum strain re- the bottom gauge (µɛ) (Absolute value); and ɛT = strain in the top
sponses for Path P2 Run 1 and Path P3 Run 1, respectively. gauge (µɛ) (absolute value).

© ASCE 04022097-8 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

(a)

(b)

Fig. 9. 2-D Contour plots: (a) rebar cover from GPR; and (b) possible delamination from IE.

Fig. 11 shows selected girders’ neutral axis depths for Path P1 response of the girders for the different paths and runs. Similar to
Run 1, Path P1 Run 2, Path P2 Run 1, and Path P3 Run 1. The Path P1 from the initial load test, the strain response of Girders
runs from the other paths were ignored since they had either too B1 and B2 was maximum since the truck tires were near these gird-
much noise or the strain data were too scattered to be considered. ers [Figs. 12(a and b)]. The strains across other girders gradually
The strain gauge data from the follow-up load test were ana- decreased with an increase in distance from the truck tires. Because
lyzed similarly. The strain gauge data at the midspans of the girders Run 2 was a repetition of Run 1 from Path P1, Figs. 12(a and b)
exhibited a lot of noise and vibration like the initial load test. This yielded identical results. Figs. 12(c–e) showed similar trends to
was addressed by calculating the moving average between 10 data the initial load test results. However, the strain values from the ini-
points similar to the initial load test. Figs. 12(a–e) show the strain tial load test were higher than those from the follow-up load test.

© ASCE 04022097-9 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

(c) (d)

(e)

Fig. 10. Strain versus time plots: (a) Path P1 run 1; (b) Path P1 run 2; (c) Path P2 run 1; (d) Path P3 run 1; and (e) stop location test.

For instance, the values from the initial and follow-up load tests Comparisons of the NA locations between the initial and the
were approximately 38 [Fig. 10(a)] and 26 [Fig. 12(a)], respec- follow-up load tests and the theoretical values assuming full and
tively, for strain Gauge 2B1B. This was primary due to the slightly no composite deck–girder interactions are provided in Table 4.
heavier truck used in the initial load test. Full composite action indicates a total interaction between the
Fig. 13 shows the location of the girder neutral axis from the deck and the girders with a common NA. In contrast, partial-
bottom for different paths and runs of the follow-up load test. composite action denotes the end slip between the superstructure

© ASCE 04022097-10 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


Girder B2 exhibited deviation, i.e., 90–95 mm and 120–125 mm
for the initial and follow-up load tests, respectively. This variation
could be due to a vibration-induced high noise-to-signal ratio of the
data, deviations of the trucks from the load paths, inconsistent
speed of the test vehicle, and temperature and/or humidity differ-
ences during the two load tests. It may be noted that the temperature
variation affects the resistance of the strain gauge and the connect-
ing wires, while humidity changes the mechanical properties of the
gauges and the adhesive layers (Lin et al. 2019; Khaled and
Mahmoud 2018). The depth of NA for a single girder was found
by averaging the depths from all load test paths, except those
where the trucks were too far from the girder to create a significant
Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

response, and there was a considerable difference between the two


load test results. To improve the accuracy of the location of the
mean NA, highly scattered data were ignored. It can be observed
from Table 4 that there is a 50 mm discrepancy between the aver-
age NA locations from the initial and follow-up load tests. How-
ever, the instrumentation of additional girders during the
follow-up test yielded more reliable results. The follow-up load
test revealed only 59% deck–girder composite action, considering
1.45 and 0.53 m theoretically calculated NA depths for full and no
Fig. 11. Girders’ neutral axis depths from the bottom. composite action, respectively. The IE results also supported this
finding by showing severe delamination on top of the girders.
elements. The NA location for a partial-composite girder will be Thus, it can be inferred that the excessive vibration of the bridge
somewhere between the NA for a full-composite and non- was due to reduced stiffness of the superstructure induced by de-
composite girder. The theoretical NA depths of the girder alone lamination in the deck and partial-composite action between the
and the composite section were calculated using Eq. (2) (AASHTO girders and the deck/panel system.
2017). For the noncomposite girder, a T-section behavior was as-
sumed since the theoretical NA was located on the web of the Load Rating
girder. The width of the compression face was taken equal to the
width of the top flange of the girder, and the distance from the ex- The following two cases were considered for the load rating:
treme compression fiber to the steel centroid was calculated from 1. Deck capacity in the transverse direction.
the top of the girder. The NA calculation for the composite section 2. Girder capacity in the longitudinal direction.
involved considering the width of the compression face of the
member equal to the effective width of the deck–girder composite
section and taking the distance from the extreme compression fiber Case 1: Deck Capacity
to the steel centroid from the top of the deck. Because the theoret- The flexural capacity of the deck was calculated using the GPR and
ical NA for the full-composite section was located on the deck, IE data following the procedure outlined here. The calculation con-
rectangular section behavior was assumed. The girder’s mild sidered 305 mm wide and 2.74 m long strips. The negative moment
steel contribution was ignored for both cases since the prestressing region was found to be 0.76 m from both sides of the centerline of
strands provide the required primary flexural resistance for typical the girder considering the contraflexure points of a continuous deck
prestressed concrete I-girder bridges. system. The GPR data were used to estimate the concrete covers for
the top rebars near the girder lines. Fig. 14 shows the analyzed neg-
A ps f pu − α1 fc′ (b − bw )hf
c= (T − section behavior) ative moment region (hatched area) on a portion of GPR data. More
f pu
α1 fc′ βbw + kA ps than 85% of the negative moment region showed a top cover rang-
dp ing from 66 to 71 mm. Therefore, an average of 68.5 mm cover was
(2)
A ps f pu used to calculate the negative flexural capacity using the following
c= (Rectangular section behavior)
f pu equation (AASHTO 2017):
α1 fc′ βb + kA ps
dp
where c = location of the theoretical NA from top of the deck/  a
girder; Aps = area of prestressing steel; fpu = specified tensile Mn− = As fy d − (3)
2
strength of prestressing steel; α1 = stress block factor; fc′ = com-
pressive strength of concrete; β = stress block factor; b = width of
the compression face of the member; bw = width of the girder where Mn− = nominal negative moment capacity; As = area of the
web; hf = depth of the top compression flange; k = 0.28 (for low re- mild steel; fy = yield stress of mild steel; d = effective depth; and
laxation steel); and dp = distance from the extreme compression a = depth of equivalent rectangular stress block.
fiber to the steel centroid. The GPR data were not useful in calculating the deck positive
It can be observed from Figs. 11 and 13 that some of the girders moment capacity since the deck bottom was not scanned. The pre-
showed similar NA locations from the initial and follow-up load cast prestressed concrete panels mainly contribute to the positive
tests, while others showed deviations. For instance, the NA loca- deck moment capacity. The positive moment region was found to
tions for Girder B1 were 100–105 mm and 105–110 mm for the ini- be 1.23 m considering the contraflexure points of a continuous
tial and follow-up load tests, respectively. Similarly, Girder B5 deck system, and the related capacity was calculated using the fol-
showed comparable results from both tests. On the other hand, lowing equation (AASHTO 2017):

© ASCE 04022097-11 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

(c) (d)

(e)

Fig. 12. Strain versus time plots, follow-up load test: (a) Path P1 run 1; (b) Path P1 run 2; (c) Path P2 run 1; (d) Path P3 run 1; and (e) stop location
test.

 a The percentage delaminations of the negative and the positive


Mn+ = A ps f ps dp − (4) moment regions were found from the 2-D IE contour plots
2
[Fig. 15] employing the procedure outlined in the “Load Rating
where Mn+ = nominal positive moment capacity; and fps = average Methodology” section. The former has 49%, while the latter has
stress in prestressing steel at nominal bending resistance = 11% delamination. It should be noted that the 2-D IE scans cannot
fpu(1 − k(c/dp)). estimate the depth of concrete delamination. To be conservative, an

© ASCE 04022097-12 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


average value of 30% delamination considering both regions is applied Table 5 summarizes the moment capacities of the deck using the
to reduce both the positive and the negative moment capacities. For GPR and IE data.
instance, the deck theoretically calculated positive moment capacity Although TxDOT allows both the Load and Resistance Factor
according to Eq. (4) was 158 kN·m/m. Considering the delamination Design (LRFD) and Load Factor Methods (LF) for bridge rating,
in the deck, the capacity was reduced by 30% to 110 kN · m/m. the latter was used herein. According to the LF Method (AASHTO
2018), the moment rating factor of the concrete components can be
calculated using the following equation:

Deck Moment Rating Factor, DRF


Moment capacity − A1 × Dead load moment
= (5)
A2 × (Live load moment + impact)
Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

where A1 = 1.3, A2 = 2.17 for inventory level, and 1.3 for operat-
ing level.
The moment capacities used in the load rating were taken from
Table 5 (modified moment capacity) and the negative flexural ca-
pacity governed the rating. The live load moment was determined
using HS-20 loading, including an impact factor of 1.3.
The bridge member ratings were calculated using Eq. (6) where
the rating factors were multiplied by the weight of the HS-20 vehi-
cle to find the allowable safe live load (inventory level) and the
maximum permissible live load (operating level). Table 6 shows
the summary of the deck moment rating factor and bridge member
rating.

Fig. 13. Girders’ neutral axis depths from the bottom, follow-up load Member rating = W × RF (6)
test.

where W = weight of the rating vehicle.


Table 4. Neutral axes locations Because the governing deck rating for the inventory level was
320 kN, it can be inferred that the deck is safe to carry HS-20
From the bottom of the girder Location (m)
load even with existing concrete delamination. However, due to se-
Theoretical full composite 1.45 vere delamination, the serviceability and the durability of the deck
Theoretical noncomposite 0.53 is of concern. During the follow-up load test, a few new cracks and
Average from the initial load test 1.02 potholes were observed on the deck, as shown in Fig. 16, which re-
Average from the follow-up load test 1.07
inforces the aforementioned statement.

Fig. 14. Negative moment region (hatched area) on a portion of the GPR contour.

(a) (b)

Fig. 15. IE contour: (a) negative moment region (hatched area); and (b) positive moment region (hatched area).

© ASCE 04022097-13 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


Case 2: Girder Capacity equation (AASHTO 2018):
The load rating of the girders was determined based on the 50
follow-up load test using Eq. (7) (AASHTO 2018). The average I= ≤ 0.3 (9)
L + 125
depth of NA from the follow-up load test was used to calculate
the girder capacity using the following equation (AASHTO 2017): where I = live load impact factor; and L = span length (in ft.;
one ft. = 0.3048 m).
Girder Rating Factor, GRF A comparison of the ratings employing the load test results as
against theoretical values assuming full-composite action between
Girder capacity − A1 × Dead load Moment
= (7) the girders and the deck/panel system is provided in Table 7. Be-
A2 × Live load Moment (1 + Impact) cause the governing girder rating for an inventory level is
490 kN, it can be concluded that the girders are still safe to carry
 
Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

 HS-20 load even after considering deck concrete delamination


a a hf
Mn = A ps f ps dp − + 0.85 fc′(b − bw ) hf − (8) and partial-composite action.
2 2 2

where Mn = nominal flexural resistance of the girder at midspan;


k = 0.28 (for low relaxation steel); c = average location of the Conclusion and Recommendations
NA from the follow-up load test (from top of the deck); b = effec-
tive width of the flanged section; bw = width of the girder top The following conclusions can be made based on the results of the
flange; and hf = depth of the top flange. study:
The dead load moment was calculated considering the self- 1. The current research employed combined NDE and static load
weight of the girders, slab, haunch, and barriers. The live load testing to diagnose a prestressed concrete girder bridge with ex-
was amplified by a multiple presence factor using the following cessive vibration. The NDE involved GPR and IE scanning of
the top of the deck to determine the reinforcement cover and lo-
cation and delamination in the deck, respectively. Two diagnos-
Table 5. Deck moment capacity
tic load tests were conducted one year apart to assess the
Moment capacity Average Modified moment condition and rate of deterioration in the test bridge. The
considering GPR percentage capacity considering study then proposed a novel load rating methodology incorpo-
Moment data (kN · m/m) delamination IE data (kN · m/m) rating the outputs from the tests, unlike the industry-wide con-
Positive 158 30 110 ventional code-based load rating practice that heavily relies on
Negative 85 30 60 information from as-built plans. The proposed method yields
more practical ratings and represents the current conditions of
the bridges more accurately than the conventional load rating
Table 6. Deck moment rating factor and bridge member rating method since it considers the existing actual rebar cover, deck
Rating level RF Bridge member rating (kN) delamination, and loss of composite action in the deck–panel
system.
Inventory 1 320
Operating 1.67 535 2. GPR scans of the bridge deck showed satisfactory top reinforce-
ment cover mostly conforming to the as-built plans, indicating

(a) (b)

Fig. 16. New pothole on the deck: (a) view from a distance; and (b) close-up view.

Table 7. Girder rating assuming partial and full-composite actions


Rating Rating factor assuming Rating assuming Rating factor based on load test Rating based on load test
level full-composite action full-composite action (kN) (partial-composite action) (partial-composite action) (kN)
Inventory 1.78 570 1.53 490
Operating 2.98 954 2.55 817

© ASCE 04022097-14 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


minimal construction error and little or no surface wearing. IE Alampalli, S. 2001. Correlation between bridge vibration and bridge deck
scanning of the top of the deck revealed severe delamination cracking: A qualitative study. Rep. No. FHWA/NY/SR-01/136.
atop of the girder lines and at the CIP-precast deck panel inter- Washington, DC: Federal Highway Administration. Special report
face, indicating loss of structural integrity and stiffness of the 136, Transportation Research and Development Bureau, New York
superstructure. The delamination was more pronounced on the State Department of Transportation, George E. Pataki, Governor/
Joseph H. Boardman, Commissioner.
top of the girders with all CIP deck concrete than away from
Almomani, Y., N. Yazdani, and E. Beneberu. 2020. “In-situ evaluation of
the girders. The average location of the NA of the composite FRP strengthening for corrosion deteriorated bridge bent caps.” J.
girders from both load tests was on the web, which indicated Bridge Eng. 25 (5): 04020019. https://doi.org/10.1061/(ASCE)BE
a significant loss of composite action between the girders and .1943-5592.0001541.
the deck/precast panel system. Based on the follow-up load ASCE. 2021. Infrastructure report card. Reston, VA: ASCE.
test, only 59% composite action in the girders existed. Together Awall, M. R., T. Hayashikawa, T. Matsumoto, and X. He. 2012. “Human
with the IE results, this leads to the conclusion that the unusual response to traffic-induced vibration of horizontally curved twin
Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

vibration in SH 75 SB Bridge over Wilson Creek was caused by I-girder bridges.” In International Conference on Noise and Vibration
the loss of composite action between the girders and the deck/ Engineering. (ISMA2012-USD2012). Leuven, Belgium: Katholieke
panel system and also the concrete delamination. Universiteit Leuven.
3. Average percentage concrete delamination obtained from the IE Barker, M. G., and K. E. Barth. 2013. “Improved serviceability criteria for
steel girder bridges.” J. Bridge Eng. 18 (7): 673–677. https://doi.org/10
scans and the actual reinforcement cover found through GPR
.1061/(ASCE)BE.1943-5592.0000402.
scanning were applied to determine realistic deck negative flex- Barker, M. G., and J. Staebler. 2011. Serviceability limits and economical
ural capacity. Furthermore, the experimental average NA was steel bridge design. Publication No. FHWA-HIF-11-044. Washington,
used to calculate the girder capacity. Load rating incorporating DC: Federal Highway Administration, US Department of Transportation.
the results of the tests showed that both the deck and the girders Bettis, G. A. 2020. Bridge inspection manual. Austin, TX: Texas
were capable of carrying HS-20 live load. However, due to se- Department of Transportation.
vere delamination of the deck and partial-composite action be- Bujnakova, P., J. Jost, and M. Farbak. 2018. “Load testing of highway brid-
tween the superstructure elements, the serviceability and ges.” In MATEC Web of Conferences 196, 02020, XXVII R-S-P Seminar
durability of the bridge are of concern. 2018, Theoretical Foundation of Civil Engineering. DOI:10.1051/ma-
The following recommendations can be made based on the study: tecconf/201819602020.
1. The proposed method used in this study may be effectively uti- Fountain, R. S., and C. E. Thunman. 1987. “Deflection criteria for steel
highway bridges.” In Proc., of the AISC National Engineering
lized for other concrete bridges with serviceability issues and
Conference in New Orleans, 1–20. Chicago, IL: American Institute of
concrete bridges in general. Steel Construction.
2. A Finite-Element Model of the bridge may be developed and Gaunt, J. T., and C. D. Sutton. 1981. Highway bridge vibration studies,
calibrated with the experimental data to conduct a parametric Joint Highway Research Project. Rep. No. JHRP-81-11. West
study that can be used to evaluate other bridges with similar Lafayette, IN: Purdue University & Indiana State Highway
conditions. Commission.
3. The levels of vibration in similar bridges may be monitored on- Goulias, D., M., and Adojam, L. S. 2015. Effective implementation of
site over time to track the age-related change that may indicate Ground Penetrating Radar (GPR) for condition assessment & monitor-
stiffness loss. ing of critical infrastructure components of bridges and highways.
4. Effective retrofitting methods to address such serviceability is- Project number SP309B4R. Hanover, MD: State Highway
sues may be developed, including considerations for relative Administration Research Report, Maryland Department of
benefit and cost. Transportation.
GPR-SLICE. 2018. GPR-SLICE v7.0 user’s manual. Woodland Hills, CA:
Geophysical Archaeometry Laboratory Inc.
Gucunski, N., F. A. Romero, P. Shokouhi, and J. Makresias. 2005.
Data Availability Statement “Complementary impact echo and ground penetrating radar evaluation
of bridge decks on I-84 interchange in Connecticut.” In Earthquake
All data, models, and code generated or used during the study Engineering and Soil Dynamics, Geotechnical Special Publication
appear in the published article. 133, edited by R. W. Boulanger, M. Dewoolker, N. Gucunski, C.
Hsein Juang, M. E. Kalinski, S. L. Kramer, M. Manzari, and J.
Pauschke, 1–10. Reston, VA: ASCE.
Gucunski, N., G. Slabaugh, Z. Wang, T. Fang, and A. Maher. 2008.
Acknowledgments “Impact echo data from bridge deck testing: Visualization and interpre-
tation.” Transp. Res. Rec. 2050 (1): 111–121. https://doi.org/10.3141
The study was performed under a grant from the Texas Department /2050-11.
of Transportation (TxDOT). The following undergraduate students Gucunski, N., Z. Wang, T. Fang, and A. Maher. 2009. “Rapid bridge deck
from the university assisted in the load test and the NDE: Sean condition assessment using three-dimensional visualization of impact
Natelli, Jose Levya, and Adams Siphanh. Support in the form of echo data.” In Proc. NDTCE’09, Non-Destructive Testing in Civil
equipment, computer, and other resources from the university are Engineering Nantes. Laboratoire Central des Ponts et Chauss’es
gratefully acknowledged. (LCPC).
Hag-Elsafi, O., and J. Kunin. 2006. Load testing for bridge rating: Route
22 over Swamp River. Albany, NY: New York State Department of
Transportation.
References Hasan, M. I., and N. Yazdani. 2014. “Ground penetrating radar utilization
in exploring inadequate concrete covers in a new bridge deck.” Case
AASHTO. 2017. Load resistance and factor design. Bridge design speci- Stud. Constr. Mater. 1: 104–114. https://doi.org/10.1016/j.cscm.2014
fications. 8th ed. Washington, DC: AASHTO. .04.003.
AASHTO. 2018. Manual for bridge evaluation, 3rd edition, with 2011, Hasan, M. I., and N. Yazdani. 2015. “Experimental study for quantitative
2013, 2014, 2015, and 2016 interim revisions. C3. Washington, DC: estimation of rebar corrosion in concrete using ground penetrating
AASHTO. radar.” J. Eng. 2016: 8536850.

© ASCE 04022097-15 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097


Image Color Extract. 2016. Accessed January 1, 2021. http://www Louisa county, Virginia. Charlottesville, VA: Virginia Transportation
.coolphptools.com/dynamic_css/doc. Research Council.
Impact Echo Instruments. 2001. A self-teaching course and reference for Matta, F., F. Bastianini, N. Galati, P. Casedei, and A. Nanni. 2005. In-site
the impact-echo method and software. Version 2.2a. Ithaca, NY: load testing of bridge A6358 (Part 1) and of bridge A6101 and bridge
Impact Echo Instruments, LLC. A6102 (Part 2). Jefferson City, MO: MODOT Research Development
Kee, S.-H., T. Oh, J. S. Popvics, R. W. Arndt, and J. Zhu. 2012. and Technology.
“Nondestructive bridge deck testing with Air-coupled impact-echo National Bridge Inventory. 2017. US department of transportation.
and infrared thermography.” J. Bridge Eng. 17 (6): 928–939. https:// Washington, DC: Federal Highway Administration.
doi.org/10.1061/(ASCE)BE.1943-5592.0000350. Park, Y. S., D. K. Shin, and T. J. Chung. 2005. “Influence of road surface
Khaled, K. M., and G. M. Mahmoud. 2018. “An investigation on the effect of roughness on dynamic impact factor of bridge by full-scale dynamic test-
humidity on the zero signal of a strain gauge measuring system.” Univ. ing.” Can. J. Civ. Eng. 32 (5): 825–829. https://doi.org/10.1139/l05-040.
Sci. 23 (1): 129–139. https://doi.org/10.11144/Javeriana.SC23-1.aint. RADAN. 2018. RADAN 7 manual. MN43-199 Rev G. Nashua, NH:
Kou, J.-W., and J. T. Dewolf. 1997. “Vibrational behavior of continuous Geophysical Survey Systems, Inc.
Downloaded from ascelibrary.org by Universidad Nacional De Ingenieria on 11/26/23. Copyright ASCE. For personal use only; all rights reserved.

span highway bridge—influencing variables.” J. Struct. Eng. 123 (3): Raju, R. K., I. Hasan, and N. Yazdani. 2018. “Quantitative relationship in-
333–344. https://doi.org/10.1061/(ASCE)0733-9445(1997)123:3(333). volving reinforcing bar corrosion and ground-penetrating radar ampli-
Lichtenstein, A. G. 1995. Bridge rating through nondestructive load test- tude.” ACI Mater. J. 115 (3): 449–457.
ing. Final Report NCHRP Proj. 12–28(13)A. Washington, DC: Roeder, C. W., K. B. Barth, and A. Bergman. 2002. Improved live load de-
Transportation Research Board. flection criteria for steel bridges. Final Report NCHRP 20-07/133.
Lin, Y., Y. Liu, and Y. Chang. 2019. “An investigation of the temperature- Seattle: Univ. of Washington.
drift effect on strain measurement of concrete beams.” Appl. Sci. 9: Zaki, A., M. A. M. Johari, W. M. A. W. Hussin, and Y. Jusman. 2018.
1662. https://doi.org/10.3390/app9081662. “Experimental assessment of rebar corrosion in concrete slab using ground
Lucas, J., T. Cousins, E. Brown, M. C, S. Sharp, R. Lane, and D. S. 2004. penetrating radar (GPR).” Int. J. Corros. 2018: 5389829. https://doi.org/10
Structural load testing and flexural analysis of the route 701 bridge in .1155/2018/5389829.

© ASCE 04022097-16 J. Bridge Eng.

J. Bridge Eng., 2022, 27(10): 04022097

You might also like